https://mastodon.social/@MrLovenstein/114874448146053278

I’m mostly questioning the statement that “opt in is missing”, as I haven’t needed to opt out of anything. The only ML that’s enabled for me is something I opted into, which implies that that’s not missing.

That just removes some buttons, but AFAICS no AI would have been running if you did not toggle those settings?

What AI did you get opted into?

(As someone not particularly into AI, I’m happy to see that the majority of the features listed here are not AI, and many of them are actually useful. I love vertical tabs and tab groups.)

Yeah it’s a spectrum, which basically runs from regular browsing -> VPN -> Tor browser for regular sites -> Tor browser for .onion sites. (And note that even .onion sites don’t need to be obscure Silk Road type sites - for example, this is DuckDuckGo. That’s still a legal privacy use case.)

Presumably, if you log in to a site, you want it to know who you are, so I think that’s fine. (Where “who you are” means “that whatever you do while logged in is being done by the same person as who did other things when logged in outside of Tor”.) So no, I don’t think you need to limit it to stuff you don’t have logins for. I’d only make sure to not login/visit a site if Tor browser actively tells you that it’s insecure (which it does when a site doesn’t use HTTPS), which is pretty obvious.

Sort of, as in, the site you’re logging into will know that you’re the same person. Obviously if it’s something like Lemmy, if you post public comments then everybody else will see that it’s the same person posting them. It used to be the case that your exit node could also see quite a bit of what you were viewing, which can indeed often be linked to things you did outside of Tor, unless the website you’re connecting to was using HTTPS. Nowadays, practically every website does that, so you should be good.

That said, I am not a security person, so if you’re a journalist protecting their sources or otherwise have a serious threat model, seek expert advice.

You don’t necessarily need to use it to visit obscure onion services, you can also just use it to post on Lemmy, i.e. like a VPN, except without a VPN provider that can know which domains you connect to.

I’m Analysis Paralysis as well, and I just figure that the analysis is part of the fun of the game.

I think he did, but I’m not sure if he called that out explicitly. Basically the recommendation is: yeah, try it, but also, all the power it gives you can make you go off the deep end. Don’t fall for the trap of trying to build your own editing software.

That would be nice, except I’d be two hours late for work :( Works great in spring though.

WebKit is Safari, Firefox is Gecko. But Gecko is mainly supported by Mozilla, so if Mozilla can’t support it financially, it remains to be seen whether someone else can or will (and if so, how).

It’s a joke, not a real point, don’t worry.

Things can very much be co-related without one causing the other (e.g. when both are consequences of a third cause). And of course, correlated things can be completely unrelated still.

(And to emphasise: yes, it is also possible that there is a causal relation between correlated things.)

Yeah, as I mentioned in the other reply, I’m not saying there’s no causation. I was just annoyed by the Freakonomics book that didn’t give any reason to believe there was, other than the correlation.

Drivers are the problem.

No no no, I should be using dependency cooldowns. You all should be finding the vulnerabilities and getting them fixed before they reach me.

Right, I’m just venting my old frustration with that specific book because they only used the correlation as “proof”, rather than indeed looking at more causal signals like studies on lead poisoning.

It is certainly also true that correlation doesn’t mean that there’s no causation, even in cases were there are no other experiments yet to support a causal relationship.

The Ian know it nice, but mainly because it’s so fast. The end result I believe is the same as a balanced granny knot. The Berluti knot is even less likely to accidentally get untied, and is about as easy to intentionally untie.

Ah, is this that claim from Freakomics that they made right after explaining that correlation doesn’t imply causation?